Pith. sign in

REVIEW

VarIabiLity seLection of AstrophysIcal sources iN PTF (VILLAIN) II. Supervised classification of variable sources

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2304.09905 v1 pith:3MFVMDDL submitted 2023-04-19 astro-ph.GA astro-ph.COastro-ph.IM

VarIabiLity seLection of AstrophysIcal sources iN PTF (VILLAIN) II. Supervised classification of variable sources

classification astro-ph.GA astro-ph.COastro-ph.IM
keywords variabilitysourcesquasarsmagnitudesonlyparametersperformancesupervised
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

Context. Large, high-dimensional astronomical surveys require efficient data analysis. Automatic fitting of lightcurve variability and machine learning may assist in identification of sources including candidate quasars. Aims. We aim to classify sources from the Palomar Transient Factory (PTF) as quasars, stars or galaxies, and to examine model performance using variability and colours. We determine the added value of variability information as well as quantifying the performance when colours are not available. Methods. We use supervised learning in the form of a histogram-based gradient boosting classifier to predict spectroscopic SDSS classes using photometry. For comparison, we create models with structure function variability parameters only, magnitudes only and using all parameters. Results. We achieve highly accurate predictions for 71 million sources with lightcurves in PTF. The full model correctly identifies 92.49 % of spectroscopically confirmed quasars from the SDSS with a purity of 95.64 %. With only variability, the completeness is 34.97 % and the purity is 58.71 % for quasars. The predictions and probabilities of PTF objects belonging to each class are made available in a catalogue, VILLAIN-Cat, including magnitudes and variability parameters. Conclusions. We have developed a method for automatic and effective classification of PTF sources using magnitudes and variability. For similar supervised models, we recommend using at least 100,000 labeled objects, and we show how performance scales with data volume.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.